Processing apparatus for behavior model

Abstract

To provide a behavior model generation device that can appropriately provide a service that reflects hidden meaning on the behavior of a subject by using behavior history data of the subject.SOLUTION: A processing apparatus for a behavior model comprises: a classification model generation unit 2a that generates a classification model configured to generate attribute information for every subject from behavior history data of each of the plurality of subjects; a proposal model generation unit 2b that, for each of a plurality of items related to the attribute of the subject, generates, from viewing history data of each subject, a proposal model configured to generate a schedule for a viewing approach for a predetermined content suitable for allowing the subject to view the predetermined content; a viewing degree information generation unit 3 that, by using the viewing history data and the classification model of each of the plurality of subjects, generates viewing degree information representing the relationship between the degree of viewing of the predetermined content and the attribute of the subject; and an information output unit 4 that outputs information obtained from the proposal model and the viewing degree information generation unit 3.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a processing device for an action model for predicting the actions of a target person based on action history data.

Background Art

[0002] Conventionally, as an action prediction model of this type, as shown in Patent Document 1 below, an action prediction device includes an action history data storage unit that stores action history data of a user, and the user's action history data stored in the action history data storage unit. An action model creation unit that probabilistically creates an action model of the user from the action history data, and an action prediction unit that probabilistically predicts the action of the user according to a predetermined condition based on the action model of the user created by the action model creation unit. It is known that when the number of data of action history data that conforms to a predetermined condition is less than a predetermined threshold value, the action prediction unit expands the condition and predicts the action of the user.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, such a conventional action prediction model merely probabilistically predicts the actions of the target person from the action history data, and there is a problem that the content represented by the action history data, that is, the meaning hidden in the actions of the target is not sufficiently considered.

[0005] This is the same even when the action history data is used as teacher data for machine learning instead of probability, and there is a problem that the meaning hidden in the actions of the target person is not reflected.

[0006] The present invention has been made in view of such a background, and an object thereof is to provide a processing apparatus for an action model that can appropriately provide a service that reflects the meaning hidden in actions by using the action history data of a target person.

Means for Solving the Problems

[0007] The processing apparatus for an action model according to the first invention includes a classification model configured to generate attribute information for each target person from the action history data of each of a plurality of target persons, a proposal model configured to generate a schedule of an action approach of the predetermined content suitable for causing the action of the predetermined content from the action history data of each target person for each of a plurality of items related to the attributes of the target person, an action degree information generation unit configured to generate action degree information representing the relationship between the degree of action of the action of the predetermined content and the attributes of the target person by using the action history data of each of the plurality of target persons and the classification model, and an information output unit configured to output information obtained from the proposal model and the action degree information generation unit and is characterized by including the above.

[0008] The processing apparatus for an action model according to the second invention is characterized in that the classification model and the proposal model are models created from a natural language processing model configured to perform learning processing by using the action history data of the plurality of target persons instead of language data and generate a feature vector of the input action history data.

[0009] Note that a target person means a person having action history data associated with an ID (in addition to viewing history data of TV broadcasts, in addition to position data (including GPS data), purchase data, store visit data, event participation (seminar participation) data, WEB access history data).

[0010] In addition, as the schedule of the action approach of the predetermined content generated by the proposed model, it is a schedule for providing information for behavior modification, etc., and specifically, it includes the planning and execution schedule of information provision (including advertisements) via various media.

[0011] According to the first invention, by creating the classification model, it is possible to generate attribute information for each of a plurality of target persons from the action history data of each target person. Further, by creating the proposed model, for each of a plurality of items regarding the attributes of the target person, it is possible to generate a schedule of the action approach of the predetermined content suitable for causing the target person to perform an action of the predetermined content from the action history data of the target person.

[0012] Furthermore, since it has an action degree information generation unit, it is possible to generate action degree information representing the relationship between the action degree of the action of the predetermined content and the attributes of the target person by using the action history data of each of a plurality of target persons and the classification model.

[0013] Then, the information obtained from the proposed model and the action degree information generation unit is output by the information output unit. Based on the output information, for each of a plurality of items regarding the attributes of the target person, an appropriate schedule of the action approach of the predetermined content and action degree information representing the relationship between the action degree of the action of the predetermined content and the attributes of the target person are recognized. As a result, it becomes possible to easily obtain an index of what schedule is appropriate for performing the action approach corresponding to the attributes and items targeted by the action of the predetermined content.

[0014] Therefore, according to the first invention, it is possible to appropriately provide services related to actions by using the action history data of a plurality of target persons.

[0015] In the first invention described above, the classification model and the proposal model may adopt an aspect in which learning processing is performed using the action history data of the plurality of target persons instead of language data, and the models are created from a natural language processing model configured to generate feature vectors of the input action history data (second invention).

[0016] According to this, it becomes possible to easily construct the classification model and the proposal model by using known natural language processing models such as GPT and BERT.

Brief Description of Drawings

[0017]

Figure 1

Mode for Carrying Out the Invention

[0018] An embodiment of the present invention will be described with reference to FIG. 1. Referring to FIG. 1, a broadcast processing apparatus as an example of a processing apparatus for an action model in the present embodiment is an apparatus configured by one or more computers or one or more electronic circuit units, or a combination thereof, and as functions realized by its hardware configuration and the installed program (software configuration), it includes a basic model creation processing unit 1, a fine-tuning processing unit 2, a viewing degree information generation unit 3 (corresponding to the action degree information generation unit of the present invention), and an information output unit 4.

[0019] The basic model creation processing unit 1 is a processing unit that creates a basic model for generating a feature vector of the viewing history data from the viewing history data of an arbitrary target viewer, and the learning processing (STEP1) in the basic model creation processing unit 1 is taken as the first step.

[0020] In the present embodiment, as the basic model, for example, natural language processing models such as GPT and BERT are used. In this case, the basic model is created by performing learning processing on the natural language processing model using the viewing history data of a plurality of target viewers instead of language data.

[0021] In this learning process, the viewing history data used instead of language data is time-series data of a plurality of unit viewing periods within a predetermined period (for example, a period of 〇 months starting from 〇:〇〇 on 〇 / 〇 / 〇, a period from 〇:〇〇 on 〇 / 〇 / 〇 to 〇:〇〇 on 〇 / 〇 / 〇, etc.). Each unit viewing period is a period of a predetermined unit time width (for example, 5 minutes, etc.) during which TV broadcasts are viewed. And for each unit viewing period, its representative time (start time, end time, central time, etc.) is associated with the broadcasting station or channel of the TV broadcast broadcast during that unit viewing period, and identification data is attached to each unit viewing period.

[0022] And in the learning process of creating the basic model, the entire viewing history data of each target viewer is regarded as text data, and the identification data for each unit viewing period is regarded as a word and input into the natural language processing model. As a result, the basic model is constructed to generate a feature vector representing the features of each terminal viewing period in the viewing history data when the viewing history data of any target viewer is input.

[0023] An additional learning result verification step for verifying the learning processing result may be performed on the basic model constructed by the learning process in this way.

[0024] More specifically, in the learning result verification step, a part of the viewing history data is masked, and the process of having the basic model predict a part of the masked viewing data is performed by changing the mask positions in order to verify whether the learning processing result of the basic model has a certain correct answer rate. In this way, the learning processing of the basic model can be ensured.

[0025] Furthermore, an additional viewing prediction step for predicting whether a target viewer will perform a specific viewing may be performed on the basic model constructed by the learning process.

[0026] More specifically, in the viewing prediction step, by using the viewing history data of a target viewer before a specific viewing as teacher data and inputting it into the basic model for retraining, it is determined whether the target viewer will perform the specific viewing. As a result, the basic model can predict the specific viewing of the viewer, and by accumulating this for a plurality of viewers, predicted viewing data such as a predicted viewing rate can be generated.

[0027] The fine-tuning processing unit 2 creates two types of models (second models) obtained by fine-tuning the basic model constructed as described above. The two types of models are a classification model created by the classification model creation unit 2a and a proposed model created by the proposed model creation unit 2b, and the creation process of the classification model in the classification model creation unit 2a (STEP2a) and the creation process of the proposed model in the proposed model creation unit 2b (STEP2b) are each the second step.

[0028] The classification model is configured to generate attribute information of a target viewer from the viewing history data of any target viewer. In this case, the attribute information of the target viewer is information regarding a plurality of predetermined types of attributes. The plurality of types of attributes are, for example, information representing gender and age, information representing income, information representing occupation, information representing the composition of cohabiting family members, information representing viewing awareness, information representing purchase awareness, and the like. And the classification model is created from the basic model as a separate model for each of these plurality of types of attributes.

[0029] The classification model corresponding to each attribute is created from the basic model so that when the viewing history data of an arbitrary target viewer is input, the target viewer can specify the content of the type of attribute corresponding to the classification model. For example, the classification model corresponding to the attribute of gender and age is created so that it can specify whether the gender is male or female and which age group from the input viewing history data. The classification models corresponding to various types of attributes are created by applying machine learning using the viewing history data of the target viewer for whom the corresponding attribute has been determined as teacher data.

[0030] The proposed model is created from the basic model so as to generate a schedule of a broadcast of predetermined content suitable for causing each target viewer to view the broadcast of the predetermined content from the viewing history data of each target viewer for each of a plurality of target items regarding the attributes of the target viewers. In this case, the target item is the content of the attributes of the target viewers who want to view the broadcast of the predetermined content. For example, the content of the attributes regarding gender and age is women aged from 〇 to 〇, and the content of the attributes regarding income is 〇 yen or more per year, and the content of the attributes regarding occupation is an office worker, etc. Such content is adopted as the target item. A plurality of such target items are set in advance.

[0031] The proposed model corresponding to each target item is created from the basic model so as to generate a schedule of a broadcast of predetermined content suitable for causing the target viewer corresponding to that target item to view the broadcast of the predetermined content when the viewing history data of the target viewer corresponding to that target item is input.

[0032] For example, when the proposed model corresponding to a certain target item attempts to cause the target viewer corresponding to that target item to view a commercial broadcast of a certain electrical product, it is created so as to be able to specify an appropriate broadcast date and time for the commercial broadcast (a broadcast date and time when the target viewer is highly likely to view the commercial broadcast. For example, between 19:00 and 19:20 every Tuesday, etc.). The proposed model corresponding to each target item is created by applying machine learning using the viewing history data of the target viewer corresponding to that target item as teacher data.

[0033] The viewing degree information generation unit 3 generates viewing degree information representing the relationship between the viewing degree of a broadcast of predetermined content and the attributes of the target viewers, using the viewing history data (viewing history data including broadcasts of predetermined content) of each of the plurality of target viewers and the classification model. More specifically, the viewing degree information indicates which attributes or target items of the target viewers have the most viewers for the broadcast of the predetermined content (in other words, which attributes or target items have the highest viewing degree (number of viewers) for the broadcast of the predetermined content).

[0034] For example, by this viewing degree information generation unit 3, viewing degree information such as that the viewing degree of a commercial broadcast of electrical products is the highest among target viewers with attributes of male aged XX to XX years old, income of more than XX yen per year, and occupation of company employee can be obtained.

[0035] The information output unit 4 outputs the information obtained by the proposal model and the viewing degree information generation unit 3 via a display (not shown) or a printer. Specifically, when the user inputs broadcast content such as a commercial broadcast scheduled for broadcast and target items, the information output unit 4 outputs to the user an appropriate broadcast schedule for making the target viewers who match the target items view the broadcast content by the proposal model. Also, when the user performs an operation to request viewing degree information for the broadcast content scheduled for broadcast, the information output unit 4 outputs the viewing degree information obtained by the viewing degree information generation unit 3 to the user.

[0036] Since the broadcast processing apparatus of the present embodiment is configured as described above, it is possible to appropriately provide services related to television viewing without requiring a great deal of man-hours of the user, using the viewing history data of television broadcasts by a plurality of viewers. Also, the classification model and the proposal model required for the service can be easily constructed using known natural language processing models.

[0037] In addition, in this embodiment, the viewing history data may be obtained not only through broadcasting (including television, radio, BS, and CS), but also through communication (including IP broadcasting and connected TVs such as Netflix (registered trademark)).

[0038] Also, in this embodiment, commercial broadcasting (video commercials) has been used as an example for explanation, but the present invention is not limited thereto. In addition to video works such as programs, movies, animations, CGs, dramas, games, and Internet distributions, radio commercials and radio programs without video, photos, novels, comics, books, music (vocal music, performances), NFTs, art works (fine arts, architecture, designs, etc.), and artistic expressions (performances) may also be used.

[0039] Also, in the above embodiment, viewing history data, which is an example of behavior history data, is used to construct the classification model and the proposal model. However, instead of the viewing history data, predetermined types of history data related to the behavior of the target person may be used.

[0040] As the predetermined types of history data (behavior history data), various history data can be adopted as long as they are data linked to the ID of the target person. For example, as the history data, in addition to location data, purchase data (such as camera purchase history data), store visit data, event participation (seminar participation) data, and web access history data may be used.

[0041] Also in the case of these various action history data, the classification model is created from the basic model as separate models for each of multiple types of attributes. Further, the proposal model is created from the basic model so as to generate a schedule of action approaches of a predetermined content suitable for causing a predetermined action from the various action history data of each target person for each of a plurality of target items regarding the attributes of the target person. For example, in the case of position data, an appropriate movement (visit) timing, in the case of purchase data, an appropriate purchase timing, in the case of store visit data, an appropriate store visit timing, in the case of event participation (seminar participation) data, an appropriate event participation timing, and in the case of web access history data, a schedule of timings such as information provision for causing an appropriate web access timing is generated. That is, it is a schedule of information provision and the like for action transformation, and specifically, a planning and execution schedule of information provision (including advertisements) via various media is generated.

[0042] Note that in the above-described embodiment, the basic model is created using natural language processing, and the classification model and the proposal model are created from the basic model. However, the classification model and the proposal model may be created without using a natural language processing model.

Explanation of Signs

[0043] 1... Basic model creation processing unit, 2... Fine-tuning processing unit, 2a... Classification model creation unit, 2b... Proposal model creation unit, 3... Audience degree information generation unit (action degree information generation unit), 4... Information output unit.

Claims

1. A classification model configured to generate attribute information for each of a plurality of subjects from the behavior history data of each of the subjects; A proposed model configured to generate a schedule of a behavioral approach having a predetermined content suitable for causing a subject to behave in a predetermined manner from the behavioral history data of each subject for each of a plurality of items related to the subject's attributes; a behavior degree information generating unit that generates behavior degree information representing a relationship between a behavior degree of the predetermined content of behavior and attributes of the target person by using the behavior history data of each of the multiple targets and the classification model; an information output unit that outputs information obtained from the proposed model and the behavior degree information generation unit; A processing device for a behavioral model comprising:

2. 2. The behavioral model processing device according to claim 1, A processing device for a behavioral model, characterized in that the classification model and the proposed model are models created from a natural language processing model configured to perform a learning process using the behavioral history data of the multiple subjects instead of language data and to generate feature vectors of the input behavioral history data.

Images